Download ip-shiv2004-routing-I

Routing I: Basic Ideas
Shivkumar Kalyanaraman
Rensselaer Polytechnic Institute
[email protected]
Based in part upon slides of Prof. Raj Jain (OSU), S. Keshav (Cornell), J. Kurose (U Mass), Noel
Chiappa (MIT)
Shivkumar Kalyanaraman
1
Overview

Routing vs Forwarding
Forwarding table vs Forwarding in simple topologies
Routers vs Bridges: review
Routing Problem
Telephony vs Internet Routing
Source-based vs Fully distributed Routing
Distance vector vs Link state routing
Addressing and Routing: Scalability

Refs: Chap 8, 11, 14, 16 in Comer textbook

Books: “Routing in Internet” by Huitema, “Interconnections” by Perlman
Reading: Notes for Protocol Design, E2e Principle, IP and Routing: In PDF
Reading: Routing 101: Notes on Routing: In PDF | In MS Word
Reading: Khanna and Zinky, The revised ARPANET routing metric
Reference: Garcia-Luna-Aceves: "Loop-free Routing Using Diffusing Computations" :
Reading: Alaettinoglu, Jacobson, Yu: "Towards Milli-Second IGP Convergence"












Shivkumar Kalyanaraman
2
Where are we?








Routing vs Forwarding
Forwarding table vs Forwarding in simple topologies
Routers vs Bridges: review
Routing Problem
Telephony vs Internet Routing
Source-based vs Fully distributed Routing
Distance vector vs Link state routing
Addressing and Routing: Scalability
Shivkumar Kalyanaraman
3
Routing vs. Forwarding

Forwarding: select an output port based on destination
address and routing table
 Data-plane function
 Often implemented in hardware

Routing: process by which routing table is built..
 … so that the series of local forwarding decisions
takes the packet to the destination with high
probability, and …(reachability condition)
 … the path chosen/resources consumed by the packet
is efficient in some sense… (optimality and filtering
condition)
Control-plane function
 Implemented in software

4
Shivkumar Kalyanaraman
Forwarding Table

Can display forwarding table using “netstat -rn”
 Sometimes called “routing table”
Destination
127.0.0.1
192.168.2.
193.55.114.
192.168.3.
224.0.0.0
default
Gateway
Flags
127.0.0.1
192.168.2.5
193.55.114.6
192.168.3.5
193.55.114.6
193.55.114.129
UH
U
U
U
U
UG
Ref
Use Interface
0 26492 lo0
2
13
fa0
3 58503
le0
2
25
qaa0
3
0
le0
0 143454
Shivkumar Kalyanaraman
5
Forwarding Table Structure







Fields: destination, gateway, flags, ...
Destination: can be a host address or a network
address. If the ‘H’ flag is set, it is the host address.
Gateway: router/next hop IP address. The ‘G’ flag says
whether the destination is directly or indirectly connected.
U flag: Is route up ?
G flag: router (indirect vs direct)
H flag: host (dest field: host or n/w address?)
Key question:
Why did we need this forwarding table in the first
place ?
Shivkumar Kalyanaraman
6
Routing in Simple Topologies
...
Bus: Drop pkt on the wire…
Full mesh: port# = dest-addr
S
Ring: send packet consistently
in (anti-)clockwise direction
Star: stubs point to hub;
hub behaves like full mesh
Shivkumar Kalyanaraman
7
Where are we?

Routing vs Forwarding
Forwarding table vs Forwarding in simple topologies

Routers vs Bridges: review

Routing Problem
Telephony vs Internet Routing
Source-based vs Fully distributed Routing
Distance vector vs Link state routing
Addressing and Routing: Scalability





Shivkumar Kalyanaraman
8


Recall… Layer 1 & 2
Layer 1:
 Hubs do not have “forwarding tables” – they simply
broadcast signals at Layer 1. No filtering.
Layer 2:
 Forwarding tables not required for simple topologies
(previous slide): simple forwarding rules suffice
 The next-hop could be functionally related to
destination address (i.e. it can be computed without
a table explicitly listing the mapping).
 This places too many restrictions on topology
and the assignment of addresses vis-à-vis ports
at intermediate nodes.
 Forwarding tables could be statically (manually)
configured once or from time-to-time.
 Does not accommodate dynamism in topology
Shivkumar Kalyanaraman
9
Recall… Layer 2
Even reasonable sized LANs cannot tolerate above
restrictions
 Bridges therefore have “L2 forwarding tables,” and use
dynamic learning algorithms to build it locally.
 Even this allows LANs to scale, by limiting
broadcasts and collisions to collision domains, and
using bridges to interconnect collision domains.
 The learning algorithm is purely local, opportunistic
and expects no addressing structure.
 Hence, bridges often may not have a forwarding
entry for a destination address (I.e. incomplete)
 In this case they resort to flooding – which may
lead to duplicates of packets seen on the wire.
 Bridges coordinate “globally” to build a spanning
tree so that flooding doesn’t go out of control.

Shivkumar Kalyanaraman
10
Recall …: Layer 3

Routers have “L3 forwarding tables,” and use a
distributed protocol to coordinate with other routers to
learn and condense a global view of the network in a
consistent and complete manner.

Routers NEVER broadcast or flood if they don’t have a
route – they “pass the buck” to another router.
 The good filtering in routers (I.e. restricting broadcast
and flooding activity to be within broadcast domains)
allows them to interconnect broadcast domains,

Routers communicate with other routers, typically
neighbors to collect an abstracted view of the network.
 In the form of distance vector or link state.
 Routers use algorithms like Dijkstra, Bellman-Ford to
compute paths with such abstracted views.
Shivkumar Kalyanaraman
11
Recall: Interconnection Devices
LAN=
Collision
Domain
Application
Transport
Network
Datalink
Physical
H H
B
H H
Gateway
Router
Bridge/Switch
Repeater/Hub
Extended LAN
=Broadcast
domain
Router
Application
Transport
Network
Datalink
Physical
Shivkumar Kalyanaraman
12
Summary so far

If topology is simple and static, routing is simple and may
not even require a forwarding table

If topology is dynamic, but filtering requirements are weak
(I.e. network need not scale), then a local heuristic setup
of forwarding table (bridging approach) suffices.
Further, if a) filtering requirements are strict,
b) optimal/efficient routing is desired, and
c) we want small forwarding tables and bounded
control traffic, then …
some kind of global communication, and smart distributed
algorithms are needed to condense global state in a
consistent, but yet complete way …

Shivkumar Kalyanaraman
13
What’s up in advanced routing?





Routers are efficient in the collection of the abstracted
view (control-plane filtering)
Routers accommodate a variety of topologies, and subnetworks in an efficient manner
Routers are organized in hierarchies to achieve
scalability; and into autonomous systems to achieve
complex policy-control over routing.
Routers then condense paths into next hops, either:
 depending upon other routers in a path to compute
next-hops in a consistent manner (fully distributed), or
 using a signaling protocol to enforce consistency.
Advanced routing algorithms support “QoS routing” and
“traffic engineering” goals like multi-path routing, sourcebased or distributed traffic splitting, fast re-route, path
protection etc.
Shivkumar Kalyanaraman
14
Where are we?

Routing vs Forwarding
Forwarding table vs Forwarding in simple topologies
Routers vs Bridges: review

Routing Problem

Telephony vs Internet Routing
Source-based vs Fully distributed Routing
Distance vector vs Link state routing
Addressing and Routing: Scalability





Shivkumar Kalyanaraman
15
Routing problem
 Collect,
process, and condense global
state into local forwarding information
 Global state
 inherently large
 dynamic
 hard to collect
 Hard issues:
 consistency, completeness, scalability
 Impact of resource needs of sessions
Shivkumar Kalyanaraman
16
Consistency



Defn: A series of independent local forwarding decisions
must lead to connectivity between any desired (source,
destination) pair in the network.
If the states are inconsistent, the network is said not to
have “converged” to steady state (I.e. is in a transient
state)
 Inconsistency leads to loops, wandering packets etc
 In general a part of the routing information may be
consistent while the rest may be inconsistent.
 Large networks => inconsistency is a scalability issue.
Consistency can be achieved in two ways:
 Fully distributed approach: a consistency criterion or
invariant across the states of adjacent nodes
 Signaled approach: the signaling protocol sets up local
forwarding information along the path.Shivkumar Kalyanaraman
17
Completeness

Defn: The network as a whole and every node has
sufficient information to be able to compute all paths.
 In general, with more complete information available
locally, routing algorithms tend to converge faster,
because the chances of inconsistency reduce.
 But this means that more distributed state must be
collected at each node and processed.
 The demand for more completeness also limits the
scalability of the algorithm.

Since both consistency and completeness pose
scalability problems, large networks have to be structured
hierarchically and abstract entire networks as a single
node.
Shivkumar Kalyanaraman
18
Design Choices …

Centralized vs. distributed routing
 Centralized is simpler, but prone to failure and
congestion
 Centralized preferred in traffic engineering scenarios
where complex optimization problems need to be solved
and where routes chosen are long-lived

Source-based (explicit) vs. hop-by-hop (fully distributed)
 Will the source-based route be signaled to fix the path
and to minimize packet header information?
 Eg: ATM, Frame-relay etc
 Or will the route be condensed and placed in each
header? Eg: IP routing option
 Intermediate: loose source route
Shivkumar Kalyanaraman
19
Design choices…
 Static
vs Dynamic Routing:
a) ‘route’ command [Static]
b) ICMP redirect message.[Static]
c) routing daemon.Eg: ‘routed’ [Dynamic,
connectionless]
d) A signaling protocol [Dynamic, virtualcircuit]
Shivkumar Kalyanaraman
20
Static vs Dynamic
Statically
Administrator
manually configures
forwarding table entries
+ More control
+ Not restricted to
destination-based
forwarding
- Doesn’t scale
- Slow to adapt to
network failures
Dynamically
Routers exchange network reachability
information using ROUTING PROTOCOLS.
Routers use this to compute best routes
+ Can rapidly adapt to changes
in network topology
+ Can be made to scale well
- Complex distributed algorithms
- Consume CPU, Bandwidth, Memory
- Debugging can be difficult
- Current protocols are destination-based
Practice : a mix of these.
Static routing mostly at the “edge”
Shivkumar Kalyanaraman
21
Example Dynamic Routing Model
Shivkumar Kalyanaraman
22
Where are we?








Routing vs Forwarding
Forwarding table vs Forwarding in simple topologies
Routers vs Bridges: review
Routing Problem
Telephony vs Internet Routing
Source-based vs Fully distributed Routing
Distance vector vs Link state routing
Addressing and Routing: Scalability
Shivkumar Kalyanaraman
23
Detour: Telephony routing




Circuit-setup is what is routed. Voice then follows route,
and claims reserved resources.
3-level hierarchy, with a fully-connected core
AT&T: 135 core switches with nearly 5 million circuits
LECs may connect to multiple cores
Shivkumar Kalyanaraman
24
Telephony Routing algorithm





If endpoints are within same CO, directly connect
If call is between COs in same LEC, use one-hop path
between COs
Otherwise send call to one of the cores
Only major decision is at toll switch
 one-hop or two-hop path to the destination toll switch.
Essence of telephony routing problem:
which two-hop path to use if one-hop path is full
(almost a static routing problem… )
Shivkumar Kalyanaraman
25
Features of telephone routing


Resource reservation aspects:
 Resource reservation is coupled with path reservation
 Connections need resources (same 64kbps)
 Signaling to reserve resources and the path
 Stable load
 Network built for voice only.
 Can predict pairwise load throughout the day
 Can choose optimal routes in advance
Technology and economic aspects:
 Extremely reliable switches
 Why? End-systems (phones) dumb because
computation was non-existent in early 1900s.
 Downtime is less than a few minutes per year =>
topology does not change dynamically
Shivkumar Kalyanaraman
26
Features of telephone routing
 Source

can learn topology and compute route
 Can assume that a chosen route is available as the
signaling proceeds through the network
 Component reliability drove system reliability and
hence acceptance of service by customers
 Simplified topology:
 Very highly connected network
 Hierarchy + full mesh at each level: simple routing
 High cost to achieve this degree of connectivity
Organizational aspects:
 Single organization controls entire core
 Afford the scale economics to build expensive network
 Collect global statistics and implement global changes
=> Source-based, signaled, simple alternate-path
routing
Shivkumar Kalyanaraman
27
Internet Routing Drivers

Technology and economic aspects:
 Internet built out of cheap, unreliable components as
an overlay on top of leased telephone infrastructure
for WAN transport.
 Cheaper components => fail more often =>
topology changes often => needs dynamic routing
 Components (including end-systems) had computation
capabilities.
 Distributed algorithms can be implemented
 Cheap overlaid inter-networks => several entities
could afford to leverage their existing (heterogeneous)
LANs and leased lines to build inter-networks.
 Led to multiple administrative “clouds” which
needed to inter-connect for global communication.
Shivkumar Kalyanaraman
28
Internet Routing Model

2 key features:
 Dynamic routing
 Intra- and Inter-AS routing, AS = locus of admin control

Internet organized as “autonomous systems” (AS).
 AS is internally connected

Interior Gateway Protocols (IGPs) within AS.
 Eg: RIP, OSPF, HELLO

Exterior Gateway Protocols (EGPs) for AS to AS routing.
 Eg: EGP, BGP-4
Shivkumar Kalyanaraman
29
Requirements for Intra-AS Routing




Should scale for the size of an AS.
 Low end: 10s of routers (small enterprise)
 High end: 1000s of routers (large ISP)
Different requirements on routing convergence after
topology changes
 Low end: can tolerate some connectivity disruptions
 High end: fast convergence essential to business
(making money on transport)
Operational/Admin/Management (OAM) Complexity
 Low end: simple, self-configuring
 High end: Self-configuring, but operator hooks for
control
Traffic engineering capabilities: high end only
Shivkumar Kalyanaraman
30
Requirements for Inter-AS Routing

Should scale for the size of the global Internet.
 Focus on reachability, not optimality
 Use address aggregation techniques to minimize core
routing table sizes and associated control traffic
 At the same time, it should allow flexibility in
topological structure (eg: don’t restrict to trees etc)

Allow policy-based routing between autonomous systems
 Policy refers to arbitrary preference among a menu of
available options (based upon options’ attributes)
 In the case of routing, options include advertised ASlevel routes to address prefixes
 Fully distributed routing (as opposed to a signaled
approach) is the only possibility.
 Extensible to meet the demands for newer policies.
Shivkumar Kalyanaraman
31
Intra-AS and Inter-AS routing
C.b
A.a
a
C
Gateways:
B.a
b
d
A
A.c
a
a
b
c
c
B
b
•perform inter-AS
routing amongst
themselves
•perform intra-AS
routers with other
routers in their AS
network layer
inter-AS,
intra-AS
routing in
gateway A.c
link layer
physical layer
Shivkumar Kalyanaraman
32
Intra-AS and Inter-AS routing: Example
C.b
a
Host
h1
C
b
A.a
Inter-AS
routing
between
A and B
A.c
a
d
c
b
A
Intra-AS routing
within AS A
B.a
a
c
B
Host
h2
b
Intra-AS routing
within AS B
Shivkumar Kalyanaraman
33
Basic Dynamic Routing Methods

Source-based: source gets a map of the network,
 source finds route, and either
 signals the route-setup (eg: ATM approach)
 encodes the route into packets (inefficient)

Link state routing: per-link information
 Get map of network (in terms of link states) at all
nodes and find next-hops locally.
 Maps consistent => next-hops consistent

Distance vector: per-node information
 At every node, set up distance signposts to destination
nodes (a vector)
 Setup this by peeking at neighbors’ signposts.
Shivkumar Kalyanaraman
34
Where are we?






Routing vs Forwarding
Forwarding table vs Forwarding in simple topologies
Routers vs Bridges: review
Routing Problem
Telephony vs Internet Routing
Source-based vs Fully distributed Routing

Distance vector vs Link state routing
 Bellman Ford and Dijkstra Algorithms

Addressing and Routing: Scalability
Shivkumar Kalyanaraman
35
DV & LS: consistency criterion


The subset of a shortest path is also the shortest path
between the two intermediate nodes.
Corollary:
 If the shortest path from node i to node j, with distance
D(i,j) passes through neighbor k, with link cost c(i,k),
then:
D(i,j) = c(i,k) + D(k,j)
j
i
k
Shivkumar Kalyanaraman
36
Distance Vector
DV = Set (vector) of Signposts, one for each destination
Shivkumar Kalyanaraman
37
Distance Vector (DV) Approach
Consistency Condition: D(i,j) = c(i,k) + D(k,j)
 The DV (Bellman-Ford) algorithm evaluates this recursion
iteratively.
 In the mth iteration, the consistency criterion holds,
assuming that each node sees all nodes and links mhops (or smaller) away from it (i.e. an m-hop view)
7
A
B
1
2
8
1
E
C
2
D
Example network
7
A
1
B
7
A
E
B
1
C
8
1
E
2
D
A’s 1-hop view
A’s 2-hop view
(After 1st iteration) (After 2nd Iteration)
Shivkumar Kalyanaraman
38
Distance Vector (DV)…



Initial distance values (iteration 1):
 D(i,i) = 0 ;
 D(i,k) = c(i,k) if k is a neighbor (i.e. k is one-hop
away); and
 D(i,j) = INFINITY for all other non-neighbors j.
Note that the set of values D(i,*) is a distance vector at
node i.
The algorithm also maintains a next-hop value
(forwarding table) for every destination j, initialized as:
 next-hop(i) = i;
 next-hop(k) = k if k is a neighbor, and
 next-hop(j) = UNKNOWN if j is a non-neighbor.
Shivkumar Kalyanaraman
39
Distance Vector (DV).. (Cont’d)



After every iteration each node i exchanges its distance
vectors D(i,*) with its immediate neighbors.
For any neighbor k, if c(i,k) + D(k,j) < D(i,j), then:
 D(i,j) = c(i,k) + D(k,j)
 next-hop(j) = k
After each iteration, the consistency criterion is met
 After m iterations, each node knows the shortest path
possible to any other node which is m hops or less.
 I.e. each node has an m-hop view of the network.
 The algorithm converges (self-terminating) in O(d)
iterations: d is the maximum diameter of the network.
Shivkumar Kalyanaraman
40
Distance Vector (DV) Example
A’s distance vector D(A,*):
 After Iteration 1 is:
[0, 7, INFINITY, INFINITY, 1]
 After Iteration 2 is:
[0, 7, 8, 3, 1]
 After Iteration 3 is:
[0, 7, 5, 3, 1]
 After Iteration 4 is:
[0, 6, 5, 3, 1]

7
A
B
1
2
8
1
E
C
2
D
Example network
7
A
1
B
7
A
E
B
1
C
8
1
E
2
D
A’s 1-hop view
A’s 2-hop view
(After 1st iteration) (After 2nd Iteration)
Shivkumar Kalyanaraman
41
Distance Vector: link cost changes
Link cost changes:
 node detects local link cost change
 updates distance table
 if cost change in least cost path,
notify neighbors
1
X
4
“good
Time 0 Iter. 1
Iter. 2
news
travels DV(Y) [ 4 0 1] [ 1 0 1] [ 1 0 1]
fast”
DV(Z) [ 5 1 0] [ 5 1 0] [ 2 1 0]
Y
50
1
Z
algorithm
terminates
Shivkumar Kalyanaraman
42
Distance Vector: link cost changes
Link cost changes:
 good news travels fast
 bad news travels slow “count to infinity” problem!
Time 0
Iter 1
Iter 2
60
4
X
Y
50
Iter 3
1
Z
Iter 4
DV(Y) [ 4 0 1] [ 6 0 1]
[ 6 0 1] [ 8 0 1] [ 8 0 1]
DV(Z) [ 5 1 0] [ 5 1 0]
[ 7 1 0] [ 7 1 0] [ 9 1 0]
algo
goes
on!
Shivkumar Kalyanaraman
43
Distance Vector: poisoned reverse

If Z routes through Y to get to X :
60
Y
 Z tells Y its (Z’s) distance to X is
4
infinite (so Y won’t route to X via Z) X
50
 At Time 0, DV(Z) as seen by Y is
[INF INF 0], not [5 1 0] !
Time 0
Iter 1
Iter 2
1
Iter 3
DV(Y)
[ 4 0 1] [ 60 0 1]
[ 60 0 1] [ 51 0 1]
DV(Z)
[ 5 1 0] [ 5 1 0]
[ 50 1 0] [ 7 1 0]
Z
algorithm
terminates
Shivkumar Kalyanaraman
44
Link State (LS) Approach

The link state (Dijkstra) approach is iterative, but it pivots
around destinations j, and their predecessors k = p(j)
 Observe that an alternative version of the consistency
condition holds for this case: D(i,j) = D(i,k) + c(k,j)
j
i
k

Each node i collects all link states c(*,*) first and runs the
complete Dijkstra algorithm locally.
Shivkumar Kalyanaraman
45
Link State (LS) Approach…



After each iteration, the algorithm finds a new destination
node j and a shortest path to it.
After m iterations the algorithm has explored paths, which
are m hops or smaller from node i.
 It has an m-hop view of the network just like the
distance-vector approach
The Dijkstra algorithm at node i maintains two sets:
 set N that contains nodes to which the shortest paths
have been found so far, and
 set M that contains all other nodes.
 For all nodes k, two values are maintained:
 D(i,k): current value of distance from i to k.
 p(k): the predecessor node to k on the shortest
known path from i
Shivkumar Kalyanaraman
46
Dijkstra: Initialization

Initialization:
 D(i,i) = 0 and p(i) = i;
 D(i,k) = c(i,k) and p(k) = i if k is a neighbor of I
 D(i,k) = INFINITY and p(k) = UNKNOWN if k is
not a neighbor of I
 Set N = { i }, and next-hop (i) = I
 Set M = { j | j is not i}

Initially set N has only the node i and set M has the rest
of the nodes.
At the end of the algorithm, the set N contains all the
nodes, and set M is empty

Shivkumar Kalyanaraman
47
Dijkstra: Iteration


In each iteration, a new node j is moved from set M into
the set N.
 Node j has the minimum distance among all current
nodes in M, i.e. D(i,j) = min {l  M} D(i,l).
 If multiple nodes have the same minimum distance,
any one of them is chosen as j.
 Next-hop(j) = the neighbor of i on the shortest path
 Next-hop(j) = next-hop(p(j)) if p(j) is not i
 Next-hop(j) = j
if p(j) = i
 Now, in addition, the distance values of any neighbor k
of j in set M is reset as:
 If D(i,k) < D(i,j) + c(j,k), then
D(i,k) = D(i,j) + c(j,k), and p(k) = j.
This operation is called “relaxing” the edges of node j.
Shivkumar Kalyanaraman
48
Dijkstra’s algorithm: example
Step
0
1
2
3
4
5
set N
A
AD
ADE
ADEB
ADEBC
ADEBCF
D(B),p(B) D(C),p(C) D(D),p(D) D(E),p(E) D(F),p(F)
2,A
1,A
5,A
infinity
infinity
2,A
4,D
2,D
infinity
2,A
3,E
4,E
3,E
4,E
4,E
5
2
A
B
2
1
D
3
C
3
1
5
F
1
E
2
The shortest-paths spanning tree rooted at A is called an SPF-tree
Shivkumar Kalyanaraman
49
Miscl Issues: Transient Loops



With consistent LSDBs, all
nodes compute consistent
loop-free paths
Limited by Dijkstra
computation overhead,
space requirements
Can still have transient loops
B
1
1
3
A
5
C
2
D
Packet from CA
may loop around BDC
if B knows about failure
and C & D do not
Shivkumar Kalyanaraman
50
Dijkstra’s algorithm, discussion
Algorithm complexity: n nodes
 each iteration: need to check all nodes, w, not in N
 n*(n+1)/2 comparisons: O(n**2)
 more efficient implementations possible: O(nlogn)
Oscillations possible:
 e.g., link cost = amount of carried traffic
D
1
1
0
A
0 0
C
e
1+e
e
initially
B
1
2+e
A
0
0
D 1+e 1 B
0
0
C
D
… recompute
routing
51
1
A
0 0
C
2+e
B
1+e
… recompute
2+e
A
0
D 1+e 1 B
e
0
C
… recompute
Shivkumar Kalyanaraman
Misc: How to assign the Cost Metric?





Choice of link cost defines traffic load
 Low cost = high probability link belongs to SPT and will
attract traffic
Tradeoff: convergence vs load distribution
 Avoid oscillations
 Achieve good network utilization
Static metrics (weighted hop count)
 Does not take traffic load (demand) into account.
Dynamic metrics (cost based upon queue or delay etc)
 Highly oscillatory, very hard to dampen (DARPAnet
experience)
Quasi-static metric:
 Reassign static metrics based upon overall network
load (demand matrix), assumed to be quasi-stationary
Shivkumar Kalyanaraman
52
Misc: Incremental SPF Algorithms





Dijkstra algorithm is invoked whenever a new LS update
is received.
 Most of the time, the change to the SPT is minimal, or
even nothing
If the node has visibility to a large number of prefixes,
then it may see large number of updates.
Flooding bugs further exacerbate the problem
Solution: incremental SPF algorithms which use
knowledge of current map and SPT, and process the
delta change with lower computational complexity
compared to Dijkstra
 Avg case: O(logn) compared to O(nlogn) for Dijkstra
Ref: Alaettinoglu, Jacobson, Yu, “Towards Milli-Second
Shivkumar Kalyanaraman
IGP Convergence,” Internet Draft.
53
Summary: Distributed Routing
Techniques
Link State






Vectoring
Topology information is
flooded within the routing
domain
Best end-to-end paths are
computed locally at each
router.
Best end-to-end paths
determine next-hops.
Based on minimizing some
notion of distance
Works only if policy is shared
and uniform
Examples: OSPF, IS-IS






Each router knows little
about network topology
Only best next-hops are
chosen by each router for
each destination network.
Best end-to-end paths result
from composition of all nexthop choices
Does not require any notion
of distance
Does not require uniform
policies at all routers
Examples: RIP, BGP
Shivkumar Kalyanaraman
54
Where are we?








Routing vs Forwarding
Forwarding table vs Forwarding in simple topologies
Routers vs Bridges: review
Routing Problem
Telephony vs Internet Routing
Source-based vs Fully distributed Routing
Distance vector vs Link state routing
 Bellman Ford and Dijkstra Algorithms
Addressing and Routing: Scalability
Shivkumar Kalyanaraman
55
Addressing: Objects




Address is a numerical “name” which refers to an object
There are several types of objects we’d like to refer to at
the network layer…
Interface:
 A place to which a producer or consumer of packets
connects to the network; a network attachment point
Network:
 A collection of interfaces which have some useful
relationship:
 Any interface can send directly to any other without
going through a router
 A topology aggregate
Shivkumar Kalyanaraman
56
Addressing: Objects




Route or Path:
 A path from one place in the network to another
Host:
 An actual machine which is the source or destination
of traffic, through some interface
Router:
 A device which is interconnecting various elements of
the network, and forwarding traffic
Node:
 A host or router
Shivkumar Kalyanaraman
57
Address Concept





Address: A structured name for a network interface or
topology aggregate:
The structure is used by the routing to help it scale
Topologically related items have to be given related
addresses
Topologically related addresses also:
 Allow the number of destinations tracked by the
routing to be minimized
 Allow quick location of the named interface on a map
 Provide a representation for topology distribution
 Provide a framework for the abstraction process
DNS names:
 A structured human usable name for a host, etc
 The structure facilitates the distribution and lookup
Shivkumar Kalyanaraman
58
Flat vs Structured Addresses


Flat addresses: no structure in them to facilitate scalable
routing
 Eg: IEEE 802 LAN addresses
Hierarchical addresses:
 Network part (prefix) and host part
 Helps identify direct or indirectly connected nodes
Shivkumar Kalyanaraman
59
Tradeoffs in Large Scale Routing



Tradeoff: discard detailed routing information vs incur the
overhead of large, potentially unneeded detail.
 This process is called abstraction.
There are two types of abstraction for routing:
 Compression, in which the same routing decision is
made in all cases after the abstraction as before
 Thinning, in which the routing is affected
 If the prior routing was optimal, discarding routing
information via thinning means non-optimal routes
Large-scale routing incurs two kinds of overhead cost:
 The cost of running the routing
 The cost of non-optimal routes
 Challenge of routing is managing this choice of costs.
Shivkumar Kalyanaraman
60
Hierarchical Routing Example: PNNI
Shivkumar Kalyanaraman
61
Summary
Routing Concepts
 DV and LS algorithms
 Addressing and Hierarchy

Shivkumar Kalyanaraman
62